Johnny, here are my reactions (MR) to your questions.

#1 If I only tested 'half' of this period, 2002 to say mid 2006, then the results would only be 50% accurate, because I would only have tested half of the data available.
MR: That is not what I would call 50% accurate data.

#2 If I test 'all' of the period for which the ES has been sufficently liquid 2002 until now,then the results will be 100% accurate because I will have backtested my system on 'all' of the data available.
MR: Yes the data is 100% accurate, but be careful of what conclusions you draw with it.

#3 If after testing 'half' of the period my win ratio is 60%, then I can assume that this 60% win ratio will only be right half of the time.
MR: No.

#4 If after testing my system on 'all' of the data my win ratio is 60%, then I can assume that this 60% win ratio will be right all of the time, 100% accurate.
MR: NO. A conclusion like this is an example of what I am trying to warn you against in #2.

Of course all of the above assume that the ES will continue to behave as it has in the past.
MR: Ah, there's the rub!